Access to International Capital Do the Credit Ratings Agencies Help or Hurt? Asymmetric Bias and Self-fulfilling Sovereign Defaults David Tennant, Damien King, & Marlon Tracey Paper argues that… • CRAs have reason to be biased against poor countries • Evidence suggests the bias is statistically significant • The bias can be sufficient to trigger default/restructuring Credit Rating Agency’s Objective CRAs wish to minimize both cost (of acquiring information) and inaccuracy (of their ratings). But there is a trade-off between the two, since accuracy is costly. Characteristic s of the Ratings Business • It is costly to acquire information on the ability/willingness of a sovereign to service debt • The weaker a country’s institutions, the poorer it is and also the worse is the quality of readily available information • Default by a highly rated sovereign is worse (reputationally) than failure to default by a poorly rated one 4 Accuracy costs 3 COST 2 1 UNDER-ESTIMATE OVER-ESTIMATE 0 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 COST Underestimating default likelihood worse than overestimating UNDER-ESTIMATE OVER-ESTIMATE Striking a balance = overestimating COST UNDER-ESTIMATE OVER-ESTIMATE COST Optimal overestimation worse the more costly it is to get information UNDER-ESTIMATE OVER-ESTIMATE Therefore … • CRA’s estimated probability of default has a bias, the strength of which is inversely related to a country’s level of development. • “Bias” because it is independent of the fundamentals that determine a country’s ability and willingness to Paper argues that… • CRAs have reason to be biased against poor countries • Evidence suggests the bias is statistically significant • The bias can be sufficient to trigger default/restructuring Objective of Statistical Estimation Test CRAs decision to downgrade, upgrade or leave unchanged the rating of a country’s foreign currency sovereign debt. Statistical testing takes account of… • Economic and institutional fundamentals • Debt, debt service, fiscal balance, GDP, investment, reserves, inflation, CA balance, Institutional quality • Country specific fixed effects • Some element of a country’s risk may be particular to that country, e.g., social capital • Heterogeneous thresholds • Threshold for re-grade not same for all countries • Time-period dummies • Willingness to re-grade changes over time • Tempering • General reluctance to change a rating due to desire for stability and upper/lower limits Data • Countries: 142 • Years: 1997 to 2011 • CRAs: S&P, Moody’s, Fitch Mean Ratings S&P Low Middle Upper 8.3 10.2 17.7 1.89 Moody’s 8.4 2.21 Fitch 8.3 2.26 3.26 3.14 10.3 17.8 3.25 3.22 10.4 18.0 3.37 3.10 Factors Influencing Ratings Changes S&P ∇ Debt ∇ Debt Service ∇ Real GDP per cap ∇ Investment ∇ ln Export ∇ Reserve/Import ∇ Current Account Bal. ∇ Inflation ∇ Institutional Quality -5.89 1.98 0.09 3.65 2.67 0.11 -6.84 -3.99 1.11 Moody’s Fitch -2.47 -5.76 0.83 -3.43 0.07 5.11 1.52 0.09 0.08 -2.67 -4.41 -0.45 -2.51 1.07 2.14 5.71 0.34 0.02 Factors Influencing Upgrade Thresholds S&P Low Income Middle Income Moody’s Fitch 0.62 0.73 1.14 0.08 0.22 0.32 Paper argues that… • CRAs have reason to be biased against poor countries • Evidence suggests the bias is statistically significant • The bias can be sufficient to trigger default/restructuring Government’ s Objective Governments wish to minimize both taxes and defaults. But there is a trade-off between the two since they are alternative means of financing. Characteristic s of Fiscal Choices • Defaulting is policy choice • There is a fixed cost to defaulting • CRAs can see when a government would be better off by defaulting To be or not to be (a defaulter) no default, then govt should… If CRAs expect default, then govt should… y default default not default x not default Conclusion s • Optimal for CRAs to overestimate the probability of default • Information acquisition costlier with poorer countries • Highly rated default is reputationally worse than a poorly rated survivor • Constitutes a bias • Unrelated to ability and willingness to pay. • Evidence that S&P, Moody’s, and Fitch are reluctant to upgrade poorer countries • There is a range of debt where a CRA could rationally predict either default or no default • Within that range, poor countries are more likely to get an unwarranted lower rating, which can trigger a decision to default